Osteoarthritis (OA), the most common form of arthritis, is a slowly progressing disease characterized clinically by pain, deformity, and loss of function. OA is a significant contributor to disability and loss of independence among the elderly. Knee OA is associated with a progressive reduction in function, including difficulty in changing from the sitting to the standing position, and a decrease in mobility and in the ability to carry out activities of daily living. While recent scientific advances have yielded highly effective disease-modifying therapies for rheumatoid arthritis, no such therapies exist for osteoarthritis. Because of the chronic nature of the disease and variable clinical outcomes, a better understanding of the relationship between various measures of clinically relevant or patient-centered outcomes - pain, stiffness, reduced function, use of analgesics, disability, quality of life and composite scores accounting for these and other clinically relevant outcomes are needed. Due to its chronic nature and the lack of effective disease modifying therapies a range of biobehavioral, dietary, pharmacologic and complementary and alternative (CAM) therapies are used by patients to reduce OA symptoms, pain, and improve function and quality of life. Despite their wide use, relatively few CAM therapies have been tested adequately with appropriate study designs to assess their effectiveness in improving patient-centered outcomes. The Osteoarthritis Initiative is large epidemiologic study with multiple patient-centered outcomes assessed yearly in subjects at high risk or having already developed knee OA as defined by fixed flexion radiographs at baseline. In addition at baseline, a large inventory of biobehavioral, dietary, pharmacologic and CAM therapies was assessed and most of these "treatments" are assessed either yearly or every two years throughout 8 years of projected follow-up. When studying the effectiveness of biobehavioral, dietary, pharmacologic and CAM therapies on improving patient centered outcomes related to knee osteoarthritis several analytic issues arise when evaluating epidemiological data. First no validated composite outcome measure has been developed that accounts for the full impact of knee osteoarthritis on patient-centered outcomes. While Western Ontario and McMaster Universities Osteoarthritis index (WOMAC) and Knee Injury and Osteoarthritis Outcomes Score (KOOS) summary scales have been developed they rely exclusively on subjective outcomes, and are highly inter-correlated making a summary score of questionable value. Second, participant s at baseline with worse patient-centered outcomes and knee osteoarthritis are more likely to utilize CAM therapies in an attempt to alleviate symptoms. This "confounding by indication" biases analyses of the potential beneficial of CAM therapies effects towards the null, making it difficult to discern whether they are effective for improvement of knee OA pain and symptoms. State-of-the art statistical methods have been developed that deal reasonably effectively with these analytic issues and will be employed in this project.